Test and cull

Herd-level test-and-cull programmes are directed at identifying and removing infected or infectious animals, thus reducing potential sources of infection for other susceptible livestock. This applies whether the overall aim of control is to reduce clinical disease in- cidence, reduce infection prevalence or to minimise sub-clinical losses. The term ‘paratu- berculosis’ is applied in the literature to a range of conditions including infected, infec- tious, or clinically diseased. When assessing the performance of diagnostic tests for use in test and cull programmes, it is important to be clear which of these is the target condition, as the sensitivity of diagnostic tests varies with the stage of infection and thus the target condition (Nielsen and Toft 2008).

Dependent on the aim of control, specific objectives of test and cull may include identi- fying animals likely to progress to clinical disease, identifying animals that are infected or infectious, or identifying animals that excrete a high concentration (e.g. 1000 cfu/g) of MAP in faeces. Clinically affected animals are likely to be shedding the organism in faeces (Whitlock and Buergelt 1996) and are a source of transmission via the intrauter- ine route (Whittington and Windsor 2009), so removing them from the herd will remove one identifiable source of infection. However, identifying sub-clinically infected but in- fectious animals is more challenging. Such animals may be shedding MAP before they are detectable by diagnostic tests (Collins 2003) and they thus represent one of the major limitations to the success of test-and-cull strategies that aim to reduce prevalence of in- fection.

The diagnostic tests in current use in the live animal are based either on detection of the agent, the cell-mediated immune (CMI) response, or the humoral immune response, and are considered in general below. The section concludes with a review of diagnostic tests specific to deer.

Detection of MAP

Isolation of the organism has been used as a diagnostic tool for almost 100 years. Recent advancements have included the use of radiometric culture and liquid culture systems (Collins et al 1990) to detect lower numbers of bacteria more rapidly and these techniques have also increased the sensitivity of detection of the slower-growing ovine-strain of MAP (Whittington 2010).

Intestinal tissue culture is regarded as the earliest and most sensitive means of detecting MAP infection, as it may identify non-shedders and animals with paucibacillary intestinal lesions (Perez et al 1996). Tissue culture is generally limited to necropsy specimens, but biopsy in the live animal is possible and has been used as a diagnostic technique in cattle (Pemberton 1979) and as a research tool (Mackintosh et al 2010a). However, biopsy does not have a practical application in screening programmes, so faecal culture or PCR methods are those generally applied to detect the organism in the live animal.

Sensitivity of faecal culture varies with factors such as age, culture technique and strain of organism. In dairy cattle, for example, the sensitivity of faecal culture has been shown to increase with age and lactational stress (Norton et al 2010). Liquid culture methods, such as BACTEC are more sensitive than solid culture to detect MAP, but modified BACTEC 12B is the only liquid culture medium that supports growth of all common MAP strains (Whittington 2010). Estimates of the sensitivity of faecal culture by all methods range from 23% to detect infected cattle, to 74% to detect infectious cattle (Nielsen and Toft 2008).

PCR techniques were initially considered poorly sensitive in biological samples (Grant et al 1998), but recent advances in methodology mean they are now considered to be possibly as sensitive as culture (Bolske and Herthnek 2010). PCR may be routinely applied to milk, tissue or faecal samples, and has the advantage of rapid turnaround time and MAP cell enumeration using real-time methods. It may therefore have particular application in identifying heavily shedding animals. In contrast to culture methods, there is no loss of

sensitivity when frozen faecal samples are tested (Khare et al 2008).

Detection of the cell-mediated immune response

The cell-mediated immune (CMI) response is considered to be a key determinant of im- munity to all mycobacterial infections, with the humoral response giving little or no pro- tection (Chiodini 1996, Stabel 2000). The CMI response occurs early in infection (Stabel 2000), therefore tests to detect it have the greatest potential to detect infected animals before shedding of MAP occurs.

The intradermal skin test, using avian or johnin purified protein derivative (PPD), has been used in cattle to detect CMI response to the injected antigen. Assays to detect gamma- interferon (IFN-γ) release following antigen stimulation of whole blood have also been developed (Billmanjacobe et al 1992). There has been some discussion in the literature about the value of using IFN-γ. Some authors (Jungersen et al 2002) advocate that in young animals it is a measure of exposure rather than infection, and should be used to assess the effect of control interventions, rather than to identify candidates for culling. The positive relationship in sheep between a strong CMI response and reduced intestinal pathology (Burrells et al 1998, Gwozdz et al 2000), and evidence of a correlation between negative IFN-γ test and higher ELISA OD values in cattle (Mikkelsen et al 2009), give support to that suggestion. However, others assert that in an infected herd or flock, identi- fying and removing exposed and thus potentially infected animals will ultimately reduce MAP transmission and assist control (Bosward et al 2010). The value of using IFN-γ

therefore, may be dependent on the specific aim of diagnosis and its use may be indicated more in infection eradication than disease control programmes.

Detection of the humoral immune response

While serum antibody may be detected by complement fixation (CF) or agar-gel immun- odiffusion (AGID) techniques, the indirect antibody enzyme-linked immunosorbent assay (ELISA) is the most widely used test for detection of the humoral immune response to MAP in serum or milk, and a variety of ELISA tests have been developed for use in dif- ferent ruminant species (Nielsen and Toft 2008). ELISA tests may be affected by the variability of the immune response of the individual, the stage of disease and the type of histopathological lesion i.e. paucibacillary or multi-bacillary (Clarke and Little 1996).

They have low (5-30%) sensitivity for detection of MAP-infected animals (Nielsen 2010). However, they have the advantage that they may detect infected animals before they be- come infectious, as antibody may be detectable before shedding of MAP; an important feature when the aim of test and cull is reduction of infection prevalence. Sensitivity of ELISA tests to detect infected cattle increases with increasing age, but sensitivity to de- tect infectious cattle does not appear to be age-dependent (Nielsen and Toft 2006). ELISA tests have the advantage that results can be obtained rapidly and relatively inexpensively and they are generally considered to have good sensitivity (75%) for detection of animals shedding high levels of MAP in faeces, although are less sensitive (15%) at detecting low shedders (Whitlock et al 2000).

In a recent comprehensive review of ante-mortem diagnosis of MAP, Nielsen and Toft (2008) summarised and critically reviewed ELISA and other diagnostic tests in a range of species. Target conditions were classified as ‘infected’, ‘infectious’ and ‘affected’, with ‘affected’ defined as clinically diseased or showing reduced production performance. The review identified many reports of ELISA test performance in the literature, but poor re- porting of target condition and study populations. The authors concluded that for all species there was a “profound lack of reliable test evaluations”.

Diagnostic tests relevant to deer

Limited information is available on the characteristics of diagnostic tests in deer. A study to evaluate a modified ELISA and individual faecal culture in deer (Schroen et al 2003) reported that IFC detected 67.5% (112/166) of deer with confirmed JD. However, the case definition of ‘confirmed’ appears to include faecal and tissue culture and histopathologi- cal examination and this figure is thus not a sensitivity estimate. When assessed against tissue culture as a gold standard, IFC detected 47% (44/93) of tissue culture positive deer, but identified six further tissue culture negative deer as MAP infected. However, faecal samples were frozen prior to culture, which may have reduced the subsequent recovery of bacteria (Richards 1981).

The same study found maximum sensitivity of a deer-conjugate ELISA was 51% when specificity was 59%, while maximum specificity was 99.5% when sensitivity was 36%. However, only 110 of the 172 serum samples originated from deer that were tissue cul- ture positive; the remainder were from animals classified as infected based on the results

of histopathology or faecal culture. The source of samples was recorded as “previous diagnostic submissions”, abattoirs and whole-herd on-farm testing. There were no data presented on the age of sampled deer, nor was there information on whether deer were clinically or sub-clinically affected for all analyses. All of these factors suggest that there is a possibility of bias towards samples from animals in more advanced stages of infec- tion. This, together with the small sample size for test evaluation (172 “infected” animals and 210 “non-infected”) indicates that the results should be interpreted with caution. An ELISA test used in farmed and wild deer populations in Spain (Reyes-Garcia et al 2008) appears to have been evaluated using PCR on MLN from 17 ELISA positive deer and “microbiological data” from nine ELISA negative deer and it is difficult from the in- formation provided to assess the test accuracy.

The development and estimated performance of an IgG1 ELISA test, the ParalisaT M_{, for}

deer has been described by Griffin et al. (2005). Specificity of 99.5% and sensitivity of 85% to detect infected animals was reported when the two antigens under study (PPDj and PpAg) were used in series. However, the selection process for samples for analysis of test sensitivity was not fully described. Samples of serum (number not reported) from 10 farms with a history of clinical paratuberculosis were originally submitted for ELISA testing, and the infection status of the corresponding deer was established by tissue cul- ture and histopathology. Preliminary estimates of test sensitivity were based on samples from 102 animals in which infection was confirmed. However, it is not clear whether the 102 selected comprised all confirmed cases or only a proportion, or whether selection bias may have been present, possibly influenced by knowledge of serum reactivity. There was no description of how the individual infection status of the controls or their source herds was established, other than that serum samples were sourced from herds with “no prior history or ongoing evidence of M. paratuberculosis infection”. Additionally, specificity estimation is described as using data from 508 “test-negative” animals. An estimate of sensitivity (77%) was given for detection of sub-clinically infected animals (n=250), us- ing tissue culture status as the ‘gold standard’ reference. The ELISA cut point used for this part of the analysis was not stated. A figure of 90% for sensitivity of the assay to identify infected deer with detectable pathology was based on histopathological status, although 75/150 (50%) of the histologically positive panel were tissue culture negative. The estimate of sensitivity may thus be an overestimate, as histopathology is not perfectly

sensitive and specific.

Recent advances in the application of PCR methods to New Zealand deer samples include the development of a quantitative PCR method for faecal samples (O’Brien et al 2010). The technique has the potential to identify deer shedding high numbers of MAP organ- isms for culling and thus may have direct application to control programmes which aim to reduce clinical disease and infection prevalence.

There has been no formal evaluation of the performance of individual faecal or tissue culture to detect MAP infection in the New Zealand deer population. The sensitivity and specificity of the Paralisa has not been independently validated, and the performance of the test in young naturally infected deer has not been estimated. Data from an exper- imental infection study in young deer (Mackintosh et al 2007a) found 19/68 (28%) of sub-clinically infected tissue culture positive deer to be Paralisa positive. The proportion positive was related to disease severity: 20% of deer with no visible histopathology, or “very mild non-specific lesions” were Paralisa positive compared to 100% of the clini- cally affected deer.

The effectiveness and cost-effectiveness of test-and cull strategies as a control measure for paratuberculosis are examined in more detail later in this review.